v2.4
Memory
Short Memory Overview
Session-based conversation memory with automatic expiration for AI agents.
Short Memory provides session-based conversation context for your AI agents. It stores recent messages in Redis with automatic expiration, enabling multi-turn dialogues without manually managing conversation history.
How It Works
Short Memory acts as a temporary conversation buffer. When your agent interacts with a user, each exchange is stored against a unique session identifier. Messages expire automatically based on your configured TTL, keeping only the most relevant recent context.
Key Concepts
| Concept | Description |
|---|---|
| Session | A unique conversation identified by session_id — create one per user interaction or chat thread |
| TTL (Time to Live) | How long messages persist after the last activity — ranges from 1 minute to 24 hours |
| Max Messages | Hard cap on messages per session — oldest messages evicted when limit is reached |
| Role-based messages | Each message carries a role (user or assistant) to preserve conversation structure |
Quick Start
# Add messages to a session
curl -X POST https://api.guidedmind.ai/api/v1/memory/short/messages \
-H "Content-Type: application/json" \
-H "X-Memory-Api-Key: mk_your_key_here" \
-d '{
"session_id": "session-123",
"messages": [
{
"role": "user",
"content": "What are my recent orders?"
}
]
}'
# Retrieve session messages
curl https://api.guidedmind.ai/api/v1/memory/short/messages/session-123 \
-H "X-Memory-Api-Key: mk_your_key_here"When to Use Short Memory
- Multi-turn chatbots — Maintain conversation context across exchanges
- Task-based agents — Track progress within a single workflow session
- Ephemeral interactions — When conversation history should not persist long-term
- Token budget management — Control context window size with max messages cap